Scholarlink

<p>Accurate identification of pest infestations in coconut plantations is critical for improving crop health and yield. Among various threats, whitefly infestation significantly affects leaf structure and overall plant vitality. This study proposes a Deep Learning-assisted Whitefly Detection Model (DL-WDM) that leverages aerial drone imagery for large-scale agricultural monitoring. The framework integrates image preprocessing, segmentation, and feature extraction using VGG-16, followed by a Deep Convolutional Neural Network (DCNN) for classification of coconut leaf conditions, including healthy, caterpillar-affected, drying, flaccidity, and yellowing states. The model is trained and evaluated on a dataset of 5036 images, capturing real-world variability in illumination and background conditions. Experimental results demonstrate that the proposed model achieves an accuracy of 95.71%, with overall performance, computational efficiency, disease progression analysis, and tracking scores of 92.3%, 94.72%, 93.51%, and 94.83%, respectively. These results indicate that the DL-WDM framework provides a robust and scalable solution for automated pest detection and health assessment in coconut plantations.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Whitefly detection in coconut tree leaves using deep learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月30日

查看全文
  • V. Kavithamani,
  • S. UmaMaheswari,
  • A. Johnson Santhosh

摘要

Accurate identification of pest infestations in coconut plantations is critical for improving crop health and yield. Among various threats, whitefly infestation significantly affects leaf structure and overall plant vitality. This study proposes a Deep Learning-assisted Whitefly Detection Model (DL-WDM) that leverages aerial drone imagery for large-scale agricultural monitoring. The framework integrates image preprocessing, segmentation, and feature extraction using VGG-16, followed by a Deep Convolutional Neural Network (DCNN) for classification of coconut leaf conditions, including healthy, caterpillar-affected, drying, flaccidity, and yellowing states. The model is trained and evaluated on a dataset of 5036 images, capturing real-world variability in illumination and background conditions. Experimental results demonstrate that the proposed model achieves an accuracy of 95.71%, with overall performance, computational efficiency, disease progression analysis, and tracking scores of 92.3%, 94.72%, 93.51%, and 94.83%, respectively. These results indicate that the DL-WDM framework provides a robust and scalable solution for automated pest detection and health assessment in coconut plantations.